ULTIMA: Ultimate Balance of Centralized and Distributed Benefits for Interference Management in 5G Cellular Networks

To cope with unprecedented mobile traffic explosion, various state-of-the-art wireless technologies such as small cell, mmWave, and heterogeneous networks have been intensively studied. Especially, small cell enhances the network capacity at the same area by deploying more next-generation NodeBs (gNBs) with the universal frequency reuse; nevertheless, it increases interference due to the shorter distance between cells requiring more exchange of feedback message for interference management (IM). In this paper, we first propose a practical cellular network architecture, namely EdgeSON, that provides network operators the opportunity to strike a balance between centralized and distributed cellular managements for the enhancement of the network performance with a reasonable feedback information exchange and spatio-temporal power sharing. On top of EdgeSON, we formulate an optimization problem aiming to maximize the time-averaged utility of users constrained by the time average transmit power budget per cluster. Although we exploit Lyapunov optimization to induce slot-by-slot subproblems, one of the transformed subproblems to find a set of power allocation and user scheduling is known as NP-hard. To tackle this issue, we propose a low-complex and practical interference management algorithm, namely IMPowerShare by introducing a critical user and power sharing virtual queue concepts which maximally exploit structural characteristics of EdgeSON to efficiently solve the NP-hard problem. IMPowerShare definitely differs from the existing interference management or spatio-temporal power sharing works in perspectives of practical feedback information exchange in EdgeSON and optimization framework. Finally, via extensive simulations in real gNB topologies in Korea, we verify that IMPowerShare outperforms the existing interference management algorithms in small cell networks.